1.4. Matrix Operations - math.la.asu.educheckman/S2010/242/242LN1.4.pdf1.4. Matrix Operations. 1.4....
66
1.4. Matrix Operations
Transcript of 1.4. Matrix Operations - math.la.asu.educheckman/S2010/242/242LN1.4.pdf1.4. Matrix Operations. 1.4....
1.4. Matrix Operations
1.4. Matrix Operations1.4. Matrix Arithmetic
Matrices also exist as objects in and of themselves,and it turns out we can “do things” to them, whichhave algebraic properties. This will enable us(eventually) to find another way to solve a systemof linear equations.
But first, we give some definitions, so we knowexactly what we’re talking about . . .
A =
1 2 34 5 67 8 9
10 11 12
A matrix is a rectangular arrangment of numbers.(Parentheses ( ) can be used instead of brackets [ ] togroup the entries of a matrix; however, vertical lines| | have a different meaning.)
A =
1 2 34 5 6
7 8 910 11 12
A row consists of the entries in a horizontal “slice”.(The second row of A is shown above in bold.)
A =
1 2 3
4 5 6
7 8 9
10 11 12
A column consists of the entries in a vertical “slice”.The third column of A is shown above in bold.
A row vector is a matrix with only one row, such as
[ 1 2 3 4 ]
A column vector (or just a vector) is a matrix withonly one column, such as
1234
A =
1 2 34 5 67 8 9
10 11 12
The entry in the ith row and the j column of thematrix A is denoted Ai,j (my notation) or ai,j (thebook’s notation). Thus A3,1 =?.
A =
1 2 34 5 67 8 9
10 11 12
The entry in the ith row and the j column of thematrix A is denoted Ai,j (my notation) or ai,j (thebook’s notation). Thus A3,1 = 7. “7” is sometimescalled the (3, 1)th entry of A.
A =
1 2 34 5 67 8 9
10 11 12
If the matrix A has m rows and n columns, it is saidto be “an m × n matrix”. These numbers (m × n) arealso the dimensions (or “size”) of the matrix.
The dimensions of A (above) are ???
A =
1 2 34 5 67 8 9
10 11 12
If the matrix A has m rows and n columns, it is saidto be “an m × n matrix”. These numbers (m × n) arealso the dimensions of the matrix.
The dimensions of A (above) are 4× 3.
When are two matrices equal? (When can we writeA = B?)
When are two matrices equal? (When can we writeA = B?)
When they have the same dimensions, andcorresponding entries are the same. That is, when
Ai,j = Bi,j
for “reasonable” i and j.
Now, we move on to arithmetic of matrices. Whatdo you think 1 2
3 45 6
+
0 10 10 1
is?
Now, we move on to arithmetic of matrices. Whatdo you think 1 2
3 45 6
+
0 10 10 1
is?
1 + 0 2 + 13 + 0 4 + 15 + 0 6 + 1
or
1 33 55 7
.
To be more precise, if A and B have the samedimensions, then the (i, j)th entry of A+B is Ai,j+Bi,j.(If A and B have different dimensions, we don’tbother defining A + B, and say that this expressionis unde�ned.)
We can also multiply a matrix by a real number; thisis called scalar multiplication.
−2[
1 2 00 1 2
]=
We can also multiply a matrix by a real number; thisis called scalar multiplication.
−2[
1 2 00 1 2
]=
[−2 −4 0
0 −2 −4
].
To be more precise, if r is a real number, and A isan m × n matrix, then rA is the m × n matrix whose(i, j)th entry is rAi,j.
These arithemtic operations have a lot of theexpected properties.
B + A =A + (B + C) =
r(A + B) =(r + s)A =
r(sA) =1A =
These arithemtic operations have a lot of theexpected properties.
B + A = A + B
A + (B + C) = (A + B) + C
r(A + B) = (rA) + (rB)(r + s)A = (rA) + (sA)
r(sA) = (rs)A1A = A
In normal addition, we have a special number “0”with the property that x + 0 = x for any x. Is there asimilar object for matrix addition? 1 2
−1 −13 5
+ M =
1 2−1 −1
3 5
In normal addition, we have a special number “0”with the property that x + 0 = x for any x. Is there asimilar object for matrix addition? 1 2
−1 −13 5
+
0 00 00 0
=
1 2−1 −1
3 5
This matrix is called the zero matrix, and I willdenote it by 0. (A bold zero; a normally typsetzero will be the number zero.) Note that there areactually infinitely many zero matrices, one of eachdimension.
Note that 0A = 0 for any matrix A.
Exercise: Find all ordered triples (x, y, z) of realnumbers such that[
0 x y3 2 x + y
]+ z ·
[1 0 00 1 0
]=
[1 2 03 3 2
].
Exercise: Find all ordered triples (x, y, z) of realnumbers such that[
0 x y3 2 x + y
]+ z ·
[1 0 00 1 0
]=
[1 2 03 3 2
].
Our strategy is to get an equation which looks likeA = B; then we can compare entries of the matricesto find x, y, and z. We need to use matrix arithmeticto simplify the expression on the left-hand side.
[0 x y3 2 x + y
]+ z ·
[1 0 00 1 0
]=
[1 2 03 3 2
]
[0 x y3 2 x + y
]+ z ·
[1 0 00 1 0
]=
[1 2 03 3 2
][
0 x y3 2 x + y
]+
[z 0 00 z 0
]=
[1 2 03 3 2
]
[0 x y3 2 x + y
]+ z ·
[1 0 00 1 0
]=
[1 2 03 3 2
][
0 x y3 2 x + y
]+
[z 0 00 z 0
]=
[1 2 03 3 2
][
z x y3 2 + z x + y
]=
[1 2 03 3 2
]
[0 x y3 2 x + y
]+ z ·
[1 0 00 1 0
]=
[1 2 03 3 2
][
0 x y3 2 x + y
]+
[z 0 00 z 0
]=
[1 2 03 3 2
][
z x y3 2 + z x + y
]=
[1 2 03 3 2
]Comparing entries, we get the equations z = 1, x = 2,y = 0, 3 = 3, 2 + z = 3, and x + y = 2. There isonly one solution to this system of equations, namely(x, y, z) = (2, 0, 1) .
Another example. This one is related to the system
x1 x2 x3 x4 x51000
0100
−1−2
00
0010
34
−10
∣∣∣∣∣∣∣23
−20
We described the solutions of
x1 x2 x3 x4 x51000
0100
−1−2
00
0010
34
−10
∣∣∣∣∣∣∣23
−20
using these equations:
x1 = 2 + α− 3β
x2 = 3 + 2α− 4β
x3 = α
x4 = −2 + β
x5 = β
WeBWorK will not accept the solution in this form;we need to convert them into another form.
x1 = 2 + α− 3β
x2 = 3 + 2α− 4β
x3 = α
x4 = −2 + β
x5 = β
by writing the vector
x1
x2
x3
x4
x5
in “expanded” form.
x1
x2
x3
x4
x5
=
2 + α− 3β
3 + 2α− 4βα
−2 + ββ
x1
x2
x3
x4
x5
=
2 + α− 3β
3 + 2α− 4βα
−2 + ββ
=
230
−20
+
α
2αα00
+
−3β−4β
0ββ
This “decomposition” was done by writing the vectoras the sum of three column vectors:
x1
x2
x3
x4
x5
=
2 + α− 3β3 + 2α− 4β
α�2 + β
β
=
2
3
0
�2
0
+
α
2αα00
+
−3β−4β
0ββ
This “decomposition” was done by writing the vectoras the sum of three column vectors: (1) the constantterms (using a 0 if there is no constant term),
x1
x2
x3
x4
x5
=
2 + �− 3β
3 + 2�− 4β�
−2 + ββ
=
230
−20
+
�
2�
�
0
0
+
−3β−4β
0ββ
This “decomposition” was done by writing the vectoras the sum of three column vectors: (1) the constantterms, (2) the terms involving α,
x1
x2
x3
x4
x5
=
2 + α�3�
3 + 2α�4�α
−2 + �
�
=
230
−20
+
α
2αα00
+
�3�
�4�
0
�
�
This “decomposition” was done by writing thevector as the sum of three column vectors. This“decomposition” was done by writing the vector asthe sum of three column vectors: (1) the constantterms, (2) the terms involving α, and (3) the termsinvolving β.
x1
x2
x3
x4
x5
=
2 + α−3β
3 + 2α−4βα
−2 + ββ
=
230
−20
+
α
2αα00
+
−3β−4β
0ββ
Notice that: (1) These are the only terms whichappear in the equations for the xi’s,
x1
x2
x3
x4
x5
=
2 + α−3β
3 + 2α−4βα
−2 + ββ
=
230
−20
+
α
2αα00
+
−3β−4β
0ββ
Notice that: (1) These are the only terms whichappear in the equations for the xi’s, and (2) everyterm in the last two vectors is a multiple of α or amultiple of β (so we can factor it out).
x1
x2
x3
x4
x5
=
230
−20
+ α ·
12100
+ β ·
−3−4
011
This is the form of the solution that WeBWorKaccepts.
Exercise: Find all solutions to the equation
2x− 3y + 4z = 6
and write the answer in expanded form.
Exercise: Find all solutions to the equation
2x− 3y + 4z = 6
and write the answer in expanded form.
[ 2 −3 4 | 6 ]
Exercise: Find all solutions to the equation
2x− 3y + 4z = 6
and write the answer in expanded form.
[ 2 −3 4 | 6 ] −→[
1 −32
2∣∣∣ 3
]The augmented matrix is now in reduced rowechelon form!
[1 −3
22
∣∣∣ 3]
• The free variables are y and z.
• Let y = s and z = t.
• x = 3 +32
y − 2z = 3 +32
s− 2t.
Thenxyz
=
3 +32
s− 2tst
=
300
+ s ·
3/210
+ t ·
−201
Now we move on to matrix multiplication. This is the“weird” arithmetic operation, and the question thatusually comes up in students’ minds is “Why wouldyou bother defining matrix multiplication in thiscomplicated way?” There are two good reasons; first,we haven’t really used the “shape” of the matrixyet. Second, well, this is going to take us back tosystems of linear equations. (So, until I get past theexamples, please don’t ask, “Why are we defining ABthis way?” You’ll learn everything you need to know,when you need to know it . . . )
First of all, when is the matrix product AB defined?Suppose that A is an m × n matrix and B is a p × qmatrix.
Then AB is only defined if n = p (when the numberof columns of A equals the number of rows of B), andAB is then a m× q matrix.
Quick quiz: Which of the following matrix productsdefined, and what are the dimensions of the product,where A is 2× 3, B is 3× 2, C is 2× 3, and D is 2× 4?
• AB• BA• AC• BD• DB
Quick quiz: Which of the following matrix productsdefined, and what are the dimensions of the product,where A is 2× 3, B is 3× 2, C is 2× 3, and D is 2× 4?
• AB Defined. 2× (3 = 3)× 2. Dimensions: 2× 2• BA• AC• BD• DB
Quick quiz: Which of the following matrix productsdefined, and what are the dimensions of the product,where A is 2× 3, B is 3× 2, C is 2× 3, and D is 2× 4?
• AB Defined. 2× (3 = 3)× 2. Dimensions: 2× 2• BA Defined. 3× (2 = 2)× 3. Dimensions: 3× 3• AC• BD• DB
Quick quiz: Which of the following matrix productsdefined, and what are the dimensions of the product,where A is 2× 3, B is 3× 2, C is 2× 3, and D is 2× 4?
• AB Defined. 2× (3 = 3)× 2. Dimensions: 2× 2• BA Defined. 3× (2 = 2)× 3. Dimensions: 3× 3• AC Undefined. 2× (3 6= 2)× 3.• BD• DB
Quick quiz: Which of the following matrix productsdefined, and what are the dimensions of the product,where A is 2× 3, B is 3× 2, C is 2× 3, and D is 2× 4?
• AB Defined. 2× (3 = 3)× 2. Dimensions: 2× 2• BA Defined. 3× (2 = 2)× 3. Dimensions: 3× 3• AC Undefined. 2× (3 6= 2)× 3.• BD Defined. 3× (2 = 2)× 4. Dimensions: 3× 4• DB
Quick quiz: Which of the following matrix productsdefined, and what are the dimensions of the product,where A is 2× 3, B is 3× 2, C is 2× 3, and D is 2× 4?
• AB Defined. 2× (3 = 3)× 2. Dimensions: 2× 2• BA Defined. 3× (2 = 2)× 3. Dimensions: 3× 3• AC Undefined. 2× (3 6= 2)× 3.• BD Defined. 3× (2 = 2)× 4. Dimensions: 3× 4• DB Undefined. 2× (4 6= 3)× 2.
Next, we need to determine what the entries of ABare. To do this, we use a generalized version of thedot product, which tells us how to multiply a rowvector by a column vector:
[x1 x2 · · · xk ] ·
y1
y2
...yk
= x1y1 + x2y2 + · · ·+ xkyk.
Now, I can give you the rule for finding oneparticular entry of AB: The (i, j)th entry (the one inthe ith row and jth column of AB is the dot productof the ith row of A and the jth column of B.
(Fortunately, it’s not backwards!)
Now it’s time to throw in some numbers . . .
[1 2 02 0 1
]·
1 −12 2
−1 1
=
What are the dimensions of the final answer?
[1 2 02 0 1
]·
1 −12 2
−1 1
=[ut utut ut
]
What are the dimensions of the final answer? 2� 2
Now we find the entries of the matrix . . .
[1 2 02 0 1
]·
1 −12 2
−1 1
=[ut utut ut
]
[ 1 2 0 ] ·
12
−1
= 1 · 1 + 2 · 2 + 0 · (−1) = 5
[1 2 02 0 1
]·
1 −12 2
−1 1
=[
5 utut ut
]
[ 1 2 0 ] ·
−121
= 1 · (−1) + 2 · 2 + 0 · 1 = 3
[1 2 02 0 1
]·
1 −12 2
−1 1
=[
5 3ut ut
]
[ 2 0 1 ] ·
12
−1
= 2 · 1 + 0 · 2 + 1 · (−1) = 1
[1 2 02 0 1
]·
1 −12 2
−1 1
=[
5 31 ut
]
[ 2 0 1 ] ·
−121
= 2 · (−1) + 0 · 2 + 1 · 1 = �1
[1 2 02 0 1
]·
1 −12 2
−1 1
=[
5 31 −1
]
(Note that
1 −12 2
−1 1
·[
1 2 02 0 1
]is the 3 × 3 matrix−1 1 −1
6 4 21 −2 1
.)
Matrix multiplication has the following “features”:
• AB is generally different from BA. In fact, any ofthe following can happen:
• AB is defined and BA is undefined;• AB and BA are both defined but have
different dimensions;• AB and BA are both defined, have the same
dimensions, but the entries don’t agree.
More features of matrix multiplication:
• If A =[
1 23 4
]and B =
[1 11 1
], then AB and BA
have the same dimensions, but the entries aredifferent.
• Generally, if AC = BC, then you cannot concludethat A = B. In fact, you can even have A2 = 0
with A 6= 0; for instance A =[
0 10 0
]. (A2 = AA.)
Matrix multiplication does have the followingproperties:
• A(BC) = (AB)C (provided that all multiplicationsare defined; and if one side is defined, the otheris also defined)
• A(B + C) = (AB) + (AC) (ditto)• (A + B)C = (AC) + (BC) (ditto)• r(AB) = (rA)B = A(rB) (ditto)• A0 = 0 and 0A = 0 (ditto)
Now . . . Why do we define matrix multiplication inthis bizarre way?
Let’s look at the matrix equation 1 −1 22 −2 81 −2 0
·x
yz
=
3222
Let’s multiply the left-hand side out.
x− y + 2z2x− 2y + 8zx− 2y
=
3222
This is another way of writing
x − y + 2z = 32x − 2y + 8z = 22x − 2y = 2
which is just a system of linear equations!
We can reverse the process as well; any system of lin-
ear equations can be written as a matrix equation
AX = B.
The matrix A has the coefficients of the variables, Xis the matrix of the variables, and B consists of thenumbers on the right-hand side.
Now, how do you solve the equation AX = B for X?
AX = B =⇒ X = “B
A”
(Not exactly; there are a few things we need toclarify, before we get a working formula.)
ut
